Unlocking the Potential of Dynamic Susceptibility Contrast Magnetic Resonance Perfusion for Non-Invasive Pre-Operative Diagnosis of Brain Tumors
[eng] INTRODUCTION: Dynamic Susceptibility Contrast Perfusion Magnetic Resonance Imaging (DSC-PWI) is a non-invasive imaging technique that evaluates the vascular and microvascular environments of central nervous system tissues. This technique uses the dynamic passage of a gadolinium-based contrast...
| Autor: | |
|---|---|
| Formato: | tesis doctoral |
| Estado: | Versión publicada |
| Fecha de publicación: | 2025 |
| País: | España |
| Recursos: | Universidad de Barcelona |
| Repositorio: | Dipòsit Digital de la UB |
| OAI Identifier: | oai:diposit.ub.edu:2445/223911 |
| Acesso em linha: | https://hdl.handle.net/2445/223911 http://hdl.handle.net/10803/695603 |
| Access Level: | acceso abierto |
| Palavra-chave: | Oncologia Neurociències Tumors cerebrals Ressonància magnètica Diagnòstic diferencial Oncology Neurosciences Brain tumors Magnetic resonance Differential diagnosis |
| Resumo: | [eng] INTRODUCTION: Dynamic Susceptibility Contrast Perfusion Magnetic Resonance Imaging (DSC-PWI) is a non-invasive imaging technique that evaluates the vascular and microvascular environments of central nervous system tissues. This technique uses the dynamic passage of a gadolinium-based contrast bolus to generate time-intensity curves that provide various metrics. The most commonly used metric in neuro-oncology is cerebral blood volume (CBV). DSC-PWI complements morphological MRI and provides advanced semi- quantitative information that has proven useful in the non-invasive pre-surgical differential diagnosis of brain tumors. Clinical evaluations are often visual or semi-quantitative and mainly rely on CBV. Visual assessments using color-coded maps can be insufficient in high-complexity scenarios and, in some cases, even misleading. Standard semi-quantitative analyses in routine clinical practice involve the manual selection of a region of interest (e.g., the tumor) and a reference for normalization, usually on a single axial slice. This approach can exclude large portions of the tumor from evaluation, is highly operator-dependent, and has low reproducibility. Additionally, focusing solely on CBV oversimplifies the information avail- able in the time-intensity curves, which constitute the raw data of the sequence and allow for deriving more metrics beyond CBV. Therefore, there is an unmet need in this field: on the one hand, to automate the selection of regions of interest, including the entire tumor heterogeneity, as well as the references for normalization, to make interpretations more robust and reproducible. On the other hand, to integrate multiple metrics, such as Percentage of Signal Recovery (PSR) or Peak Height (PH), beyond the isolated assessment of CBV, to extract richer and more multi- dimensional information. An interesting alternative for this latter point is to analyze the time-intensity curves as a unit of analysis in an unsupervised manner (without the need to extract predefined metrics such as CBV, PSR, or PH), an approach we have termed DSC-PWI curvology. With these approaches, the sequence would not only provide in- formation about overall vascularity, as CBV does, but also data on blood-brain barrier integrity, vascular dynamics, and aspects of cytoarchitecture and vasculoarchitecture. HYPOTHESIS: Multiparametric and/or curvological evaluation of DSC-PWI provides enhanced diagnostic capability surpassing the information offered by CBV alone. The application of three-dimensional segmentation methods, semi-automatic analyses, and voxel-level approaches reduces operator-dependence and incorporates tumor heterogeneity, improving the performance of conventional approaches based on a single metric and a single value from a manual and two-dimensional region of interest. OBJECTIVES: The objectives of this thesis are: 0. To improve the pre-surgical diagnostic precision and analysis robustness of DSC-PWI in the evaluation of brain tumors through a semi-automatic, volumetric, curvological, and multiparametric approach. 1. To differentiate types and subtypes of brain tumors in adults in a non-invasive, pre-surgical manner using these advanced DSC-PWI analysis methods. 2. To develop practical and clinically applicable classifiers that are explainable and based on a solid pathophysiological foundation. |
|---|